Modeling of Resistance Spot Welding Process Using Nonlinear Regression Analysis and Neural Network Approach on Galvanized Steel Sheet
Modeling of resistance spot welding process on galvanized steel sheet was investigated. Mathematical models developed by nonlinear multiple regression analysis and artificial neural network approach were employed in the prediction of welding quality factors, namely nugget diameter, penetration rate and tensile shear strength, under some welding conditions. According to the prediction models on quality, the prediction systems of welding process parameters were formulated respectively on the basis of Newton-Raphson iterative algorithm and cascade forward back propagation algorithm in order to obtain the desired welding quality. The results showed that the prediction precision of cascade forward back propagation algorithm was higher than Newton-Raphson iterative algorithm. The current duration had the largest prediction error, followed by electrode force and welding current. Therefore, it was concluded that the current duration was the most difficult parameter in prediction system of welding process in order to obtain the desired welding quality.
Yungang Li, Pengcheng Wang, Liqun Ai, Xiaoming Sang and Jinglong Bu
Y. Luo et al., "Modeling of Resistance Spot Welding Process Using Nonlinear Regression Analysis and Neural Network Approach on Galvanized Steel Sheet", Advanced Materials Research, Vols. 291-294, pp. 823-828, 2011